Abstract
Large-scale genomics efforts have provided the opportunity to access a comprehensive catalog of genetic alterations in multiple cancers. However, it has also become apparent that very few driver mutations are emerging and as a consequence of that, limited opportunities exist to target mutated oncogenic proteins. It is imperative therefore to develop alternative approaches to therapy that can leverage the selective vulnerabilities of tumor cells resulting from the engagement of abnormal pathway connectivity. These can be best exploited in vivo, in a context that is closer to the environment tumors strive in. To identify new relevant actionable dependencies we have developed PILOT (Patient-oriented In vivo Lethality to Optimize Treatment), a loss-of-function in vivo platform for rapid identification of potential therapeutic targets in Patient-Derived Xenografts (PDXs). By optimizing primary tumor explant and expansion, determination of tumor-initiating cell frequency trough retroviral-mediated transduction, in vivo RNA interference screens, next-generation sequencing and analytic pipelines, we have been able to establish a comprehensive “patient-centric” approach oriented towards the identification of the highest priority genetic targets in specific clinico-pathological and mutational settings. So far, the main limitation for the systematic exploitation of in vivo functional genomics systems to elucidate patient vulnerabilities in the PDX models come from the limited number of human cells contributing to tumor establishment in a transplantation setting. The frequency of these tumors initiating cells (TICs) is commonly estimated by time-consuming limiting dilution assays and may consistently vary between different tumor origins. With this in mind, we have integrated in our platform a system based on scrambled barcoded libraries that allows to directly assess the required coverage of screening libraries in each model and adjust the RNAi screens for this factor. Our coverage study demonstrated to be a powerful tool to identify the minimal number of cells/barcode required to sustain a complex library in PDX models and at the same time a step forward to personalize the in vivo screening patient-by-patient. As proof of concept, we applied our PILOT platform to systematically interrogate context-specific epigenetic dependencies in pancreatic ductal adenocarcinoma (PDAC). In addition to the well-known genetic alterations (KRAS, TP53, CDKN2A/p16, SMAD4), some epigenetic mechanisms demonstrated to play a central role in PDAC progression and some of them could intriguingly represent new points of vulnerability, due to the low-frequency of mutation (ex. collateral or synthetic lethality). Our screening system utilized fully annotated low-passage PDAC xenografts and a lentiviral library of pooled shRNAs targeting 230 potentially “druggable” epigenetic regulators adjusted for the coverage study in each PDX. Hairpin-associated molecular barcodes were quantified by massively parallel sequencing and clustered according to their depletion or enrichment in comparison to a control population before transplantation. To date, we have completed a total of 5 in vivo screens using diverse PDAC xenograft models and, applying our comprehensive mutational and functional data analytics pipeline, we have developed a high-throughput validation scheme to triage “hits” that emerge from each screen. Focusing on epigenetic regulators, we identified WDR5, a core member of the COMPASS histone H3 Lys4 (H3K4) MLL (1-4) methyltransferase complex, as a top tumor maintenance hit required across multiple PDAC tumors and associated with the presence of G1-checkpoint alterations (p53 or p16). Mechanistically, WDR5 functions to sustain proper execution of DNA replication in PDAC cells, as previously suggested by replication stress studies involving MLL1, a critical ATR substrate, and c-Myc, also found to interact with WDR5. We indeed demonstrated that the WDR5-Myc interaction is critical for this replicative function and protects the PDAC cells from the excessive DNA damage accumulation and mitotic catastrophe. Intriguingly, this checkpoint function executed by the WDR5-Myc axis to protect the S-phase seems to be an addiction of the cancer cells, that have more active replication forks to stabilize in order to sustain the abnormal proliferative burst. To confirm this new cancer-associated lethality, we demonstrated that normal cells display less sensitivity to this replication checkpoint in virtue of their proficient G1-checkpoints and reduced time spent in S-phase compared to cancer cells. So, our PILOT platform was able to illuminate new therapeutic vulnerabilities that can be rapidly evaluated in the clinic through the development of WDR5-Myc inhibitors. In the near future, we plan to extend this platform in syngeneic mouse models, where one can probe the effects of target inhibition in the context of an intact immune response and in the presence of immune checkpoint activators, and in association with approved drugs, to identify new therapeutic options for recalcitrant tumor populations.
Citation Format: Alessandro Carugo, Giannicola Genovese, Sahil Seth, Luigi Nezi, Angelo Cicalese, Daniela Bossi, Johnathon L. Rose, Andrea Viale, Luisa Lanfrancone, Timothy P. Heffernan, Giulio F. Draetta. PILOT: a patient-oriented in vivo functional platform to identify new lethalities and optimize cancer treatment. [abstract]. In: Proceedings of the AACR Special Conference: Patient-Derived Cancer Models: Present and Future Applications from Basic Science to the Clinic; Feb 11-14, 2016; New Orleans, LA. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(16_Suppl):Abstract nr B43.